NAVAL POSTGRADUATE SCHOOL · December 2007 Thesis Advisors: ... UU NSN 7540-01-280-5500 Standard...

NAVAL

POSTGRADUATE SCHOOL

MONTEREY, CALIFORNIA

THESIS

Approved for public release; distribution is unlimited

DESIGN AND PERFORMANCE EVALUATION STUDY OF A PROTOTYPE OF A TACTICAL UNMANNED AERIAL

VEHICLE

by

Teng Choon Hon Adrian

December 2007

Thesis Advisors: Kevin D. Jones Vladimir N. Dobrokhodov

THIS PAGE INTENTIONALLY LEFT BLANK

i

REPORT DOCUMENTATION PAGE Form Approved OMB No. 0704-0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instruction, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188) Washington DC 20503. 1. AGENCY USE ONLY (Leave blank)

2. REPORT DATE December 2007

3. REPORT TYPE AND DATES COVERED Master’s Thesis

4. TITLE AND SUBTITLE Design and Performance Evaluation Study of a Prototype of a Tactical Unmanned Aerial Vehicle 6. AUTHOR(S) Teng Choon Hon Adrian

5. FUNDING NUMBERS

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) Naval Postgraduate School Monterey, CA 93943-5000

8. PERFORMING ORGANIZATION REPORT NUMBER

9. SPONSORING /MONITORING AGENCY NAME(S) AND ADDRESS(ES) N/A

10. SPONSORING/MONITORING AGENCY REPORT NUMBER

11. SUPPLEMENTARY NOTES The views expressed in this thesis are those of the author and do not reflect the official policy or position of the Department of Defense or the U.S. Government. 12a. DISTRIBUTION / AVAILABILITY STATEMENT Approved for public release; distribution is unlimited

12b. DISTRIBUTION CODE

13. ABSTRACT (maximum 200 words) This thesis aims to provide a low-cost solution through integrating commercial off-the-shelf (COTS) technologies to produce a prototype of a “Tactical Unmanned Combat Aerial Vehicle UCAV” system that can be utilized by the front-line ground units in the near future. The Tactical UCAV is designed to enhance the information collection and autonomous precision strike capability of the ground units. The Tactical UCAV can also be deployed as sensor nodes as part of a larger global information grid in a network-centric warfare operation. The proposed Tactical UCAV system is comprised of a Hunter Unmanned Aerial Vehicle (HUAV), which primarily carries high resolution sensors and communication devices and is used as a mother-ship for smaller “Killer UAVs (KUAV).” The KUAV carries a mission specific set of instruments; it can be a sensor or a warhead or both depending on the desired end results. After the target is acquired by the HUAV, the target information will be transferred to the KUAV. The KUAV can then be launched in close proximity of the target with the target position update from the HUAV. This thesis will focus on the development of a prototype KUAV and the integration of the prototype with the existing HUAV “Rascal” developed and operated by the Naval Postgraduate School (NPS). The KUAV and the HUAV will form the Tactical UCAV system.

15. NUMBER OF PAGES

117

14. SUBJECT TERMS Unmanned Aerial Vehicle, UAV, Tactical UCAV, Non-Powered Glider, Commercial off-the-shelf, Simulink, LinAir, Hardware-in-the-loop, Procerus, Kestrel Autopilot, Kestrel Virtual Cockpit, Aviones 16. PRICE CODE

17. SECURITY CLASSIFICATION OF REPORT

Unclassified

18. SECURITY CLASSIFICATION OF THIS PAGE

Unclassified

19. SECURITY CLASSIFICATION OF ABSTRACT

Unclassified

20. LIMITATION OF ABSTRACT

UU NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89) Prescribed by ANSI Std. 239-18

ii


iii

Approved for public release; distribution is unlimited

DESIGN AND PERFORMANCE EVALUATION STUDY OF A PROTOTYPE OF A TACTICAL UNMANNED AERIAL VEHICLE

Teng Choon Hon Adrian

Captain, Singapore Armed Forces (Army) B.Eng (M.E.), Nanyang Technological University, 2002

Submitted in partial fulfillment of the requirements for the degree of

MASTER OF SCIENCE IN MECHANICAL ENGINEERING

from the

NAVAL POSTGRADUATE SCHOOL December 2007

Authors: Teng Choon Hon Adrian

Approved by: Kevin D. Jones, PhD

Thesis Advisor

Vladimir N. Dobrokhodov, PhD Thesis Advisor

Anthony J. Healey, PhD Chairman, Department of Mechanical and Astronautical Engineering (MAE)

iv


v

ABSTRACT

The Unmanned Aerial Vehicle (UAV) plays a critical role in the current

battlefield in the areas of information superiority, collateral damage, urban area fighting

and precision strike against high payoff targets. The desire for a shorter “kill-chain” is

driving the evolution of the Unmanned Combat Aerial Vehicle (UCAV). However, due to

high cost and limited quantities, UCAVs are currently only available to military planners

at the operational or strategic level.

This thesis aims to provide a low-cost solution through integrating commercial

off-the-shelf (COTS) technologies to produce a prototype of a “Tactical Unmanned

Combat Aerial Vehicle UCAV” system that can be utilized by the front-line ground units

in the near future. The Tactical UCAV is designed to enhance the information collection

and autonomous precision strike capability of the ground units. The Tactical UCAV can

also be deployed as sensor nodes as part of a larger global information grid in a network-

centric warfare operation.

The proposed Tactical UCAV system is comprised of a Hunter Unmanned Aerial

Vehicle (HUAV), which primarily carries high resolution sensors and communication

devices and is used as a mother-ship for smaller “Killer UAVs (KUAV).” The KUAV

carries a mission specific set of instruments; it can be a sensor or a warhead or both

depending on the desired end results. After the target is acquired by the HUAV, the target

information will be transferred to the KUAV. The KUAV can then be launched in close

proximity of the target with the target position update from the HUAV.

This thesis will focus on the development of a prototype KUAV and the

integration of the prototype with the existing HUAV “Rascal” developed and operated by

the Naval Postgraduate School (NPS). The KUAV and the HUAV will form the Tactical

UCAV system.

vi


vii

TABLE OF CONTENTS

I. INTRODUCTION........................................................................................................1 A. OPERATIONAL REQUIREMENTS............................................................1 B. PROPOSED CONCEPT OF OPERATIONS ...............................................2 C. SCOPE AND OBJECTIVES ..........................................................................4 D. THESIS ORGANIZATION............................................................................4

II. LITERATURE REVIEW ...........................................................................................7 A. EXISTING UCAV SYSTEMS........................................................................7

1. Predator B – General Atomics Aeronautical Systems, USA............7 2. Sperwer B – Sagem, France ................................................................8

B. EXISTING TACTICAL UAV SYSTEMS.....................................................8 1. Shadow 200 – AAI Corporation, USA ...............................................9 2. Watchkeeper Tactical UAV – Thales, UK.......................................10

III. BUILDING THE PROTOTYPE..............................................................................11 A. PROTOTYPE DESIGN ................................................................................11

1. Dimensions..........................................................................................12 2. Material...............................................................................................13 3. Components and Loading .................................................................13 4. Trimming and Balancing ..................................................................13 5. Completed Prototype .........................................................................14

IV. AERODYNAMICS STUDIES..................................................................................17 A. BASIC AERODYNAMICS FOR GLIDERS ..............................................17 B. SIMULATION USING ANALYTICAL TECHNIQUES ..........................18

1. Verification of Panel Codes...............................................................19 2. Geometry and Input File ...................................................................20 3. Aerodynamic Data .............................................................................21

C. ANALYTICAL TRADEOFF STUDIES......................................................21 1. Tradeoff Study 1 – L/D vs. AOA .......................................................21 2. Tradeoff Study 2 – γ vs. L/D..............................................................23 3. Tradeoff Study 3 – R vs. L/D.............................................................24 4. Tradeoff Study 4 – V vs. AOA ...........................................................25 5. Tradeoff Study 5 – E vs. V.................................................................26

V. GUIDANCE AND CONTROL.................................................................................31 A. SIMULATION USING NUMERICAL TECHNIQUES............................31

1. Analytical Results from LinAir .........................................................32 2. 6 Degree of Freedom (DOF) + Auto Pilot (AP) Simulink Model...32 3. Modeling the Prototype KUAV in the 6DOF + AP Model.............34 4. Simulation Procedures.......................................................................35 5. Results and Comparisons ..................................................................36 6. Refinement of Operational Envelope ...............................................40

viii

VI. HARDWARE-IN-THE-LOOP EXPERIMENTS...................................................43 A. ARCHITECTURE AND SETUP .................................................................43

1. Architecture........................................................................................43 2. Setup....................................................................................................45

B. TUNING AUTOPILOT PID GAINS...........................................................45 1. Aerodynamic Performance Parameters ..........................................46 2. Autopilot PID Gains ..........................................................................46

C. TEST SCENARIOS AND PROCEDURES.................................................46 1. Test Scenarios and Procedures .........................................................47

D. RESULTS AND COMPARISON.................................................................47 1. Results .................................................................................................47 2. Comparison ........................................................................................48

VII. CONCLUSIONS ........................................................................................................53 A. CONCLUSIONS ............................................................................................53 B. RECOMMENDATIONS AND FUTURE STUDIES .................................53

1. Understanding the Kestrel Autopilot ...............................................54 2. Variations to Gliding Profile.............................................................54 3. Physical Integration with “Rascal” ..................................................54 4. Data Integration with “Rascal” ........................................................54 5. Variable Payload Capability for the KUAV....................................56

APPENDIX A. TECHNICAL SPECIFICATIONS OF PREDATOR B.................57

APPENDIX B. TECHNICAL SPECIFICATIONS OF SPERWER B ...................59

APPENDIX C. TECHNICAL SPECIFICATIONS OF SHADOW 200..................61

APPENDIX D. TECHNICAL SPECIFICATIONS OF WATCHKEEPER TACTICAL UAV.......................................................................................................63

APPENDIX E. COMPONENT SPECIFICATIONS ................................................65

APPENDIX F. VERIFICATION OF PANEL CODES............................................67

APPENDIX G. INPUT FILE FOR PANEL CODE ..................................................71

APPENDIX H. DATA FILE GENERATED BY PANEL CODE............................73

APPENDIX I. INITIALIZATION FILE FOR 6DOF + AP MODEL ...................75

APPENDIX J. SIMULATION RESULTS FROM THE 6DOF + AP MODEL ....77

APPENDIX K. PROCEDURES FOR SETTING UP THE TCP/IP CONNECTION..........................................................................................................79

APPENDIX L. PROCEDURES FOR UPDATING AERODYNAMIC PERFORMANCE PARAMETERS IN AVIONES ................................................85

APPENDIX M. PID GAINS FOR PROTOTYPE ON KESTREL AUTOPILOT..87

APPENDIX N. PROCEDURES FOR RUNNING THE HIL EXPERIMENTS ....89

APPENDIX O. SIMULATION RESULTS FROM HIL ..........................................95

ix

LIST OF REFERENCES......................................................................................................97

INITIAL DISTRIBUTION LIST .........................................................................................99

x


xi

LIST OF FIGURES

Figure 1. Proposed concept of operations .........................................................................3 Figure 2. Predator B armed with Hellfire missiles............................................................7 Figure 3. Sperwer B armed with Spike LR missiles .........................................................8 Figure 4. Shadow 200 being launched from a catapult .....................................................9 Figure 5. Watchkeeper Tactical UAV.............................................................................10 Figure 6. Plan view of prototype and cross section of EH 3.0/12...................................12 Figure 7. Location of aerodynamic center and CG .........................................................14 Figure 8. Components for autonomous flight .................................................................14 Figure 9. Side and top views of the prototype KUAV ....................................................15 Figure 10. Forces acting on a glider in flight ....................................................................17 Figure 11. Geometry of glider generated by LinAir ..........................................................20 Figure 12. Tradeoff between L/D and AOA ......................................................................22 Figure 13. Tradeoff between γ, L/D and AOA...................................................................23 Figure 14. Glide path geometry.........................................................................................24 Figure 15. Tradeoff between R, L/D and AOA ..................................................................25 Figure 16. Tradeoff between V and AOA ..........................................................................26 Figure 17. Tradeoff between E, V and AOA......................................................................27 Figure 18. Comparison between R and E vs. AOA............................................................28 Figure 19. Flow chart depicting simulation process..........................................................31 Figure 20. Simulink Block Diagram for 6DOF + AP Model............................................32 Figure 21. Comparison between analytical and 6DOF + AP models for L/D vs. AOA ....37 Figure 22. Comparison between analytical and 6DOF + AP models for V vs. AOA ........38 Figure 23. Comparison between analytical and 6DOF + AP models for R vs. AOA ........39 Figure 24. Comparison between analytical and 6DOF + AP models for E vs. AOA ........40 Figure 25. HIL architecture using a TCP/IP connections .................................................44 Figure 26. HIL architecture using a standard serial modem interface ..............................44 Figure 27. HIL results for a gliding velocity of 8m/s........................................................48 Figure 28. Comparison between analytical, 6DOF + AP and HIL models for L/D vs.

AOA..................................................................................................................49 Figure 29. Comparison between analytical, 6DOF + AP and HIL models for R vs.

AOA..................................................................................................................50 Figure 30. Comparison between analytical, 6DOF + AP and HIL models for E vs.

AOA..................................................................................................................51 Figure 31. Communication protocol via ground control stations......................................55 Figure 32. Communication protocol via radio frequency module ....................................55 Figure 33. Disconnected programming wire.....................................................................79 Figure 34. Connection for the programming and power cables on autopilot....................79 Figure 35. HIL selection in Virtual Cockpit......................................................................80 Figure 36. Connection window for Aviones .....................................................................80 Figure 37. Agent added in Aviones...................................................................................81 Figure 38. HIL simulation control window.......................................................................82 Figure 39. User interfaces for Aviones and Virtual Cockpit.............................................83

xii

Figure 40. Aerodynamic performance coefficients for prototype.....................................85 Figure 41. Physics parameter editor window ....................................................................86 Figure 42. Values of PID gains .........................................................................................87 Figure 43. Flight control window of Virtual Cockpit .......................................................89 Figure 44. Launching the UAV from Aviones..................................................................90 Figure 45. UAV in “NAV’ mode ......................................................................................91 Figure 46. HIL Sim Control Window of Aviones.............................................................92 Figure 47. Data log window of Virtual Cockpit................................................................92

xiii

LIST OF TABLES

Table 1. Dimensions of prototype..................................................................................12 Table 2. Wind tunnel results of

0DC and K for wings of AR 1 and 2.............................20 Table 3. Comparison between experimental and analytical results ...............................23

xiv


xv

ACKNOWLEDGMENTS

The completion of this thesis would not be possible without Professor Vladimir

Dobrokhodov and Professor Kevin Jones.

I would like to thank Professor Vladimir Dobrokhodov for his support and

patience for the entire duration of this thesis. I am grateful for his effort in spending

endless hours explaining difficult concepts and Simulink diagrams to me in his office. He

has truly shown me what teaching is all about.

I would also like to thank Professor Kevin Jones for all his support and assistance

for the entire duration of this thesis. Without his assistance the prototype would never

have been built. I will always remember the many hours we spent in the Unmanned

System Laboratory cutting the Spyder foam and I will definitely not forget the “pleasant”

fumes and particles we had to inhale.

Finally, the advisor and student partnership will come to an end after I graduate

from NPS, but I sincerely know that the friendship we built will last for many years to

come.

xvi


1

I. INTRODUCTION

A. OPERATIONAL REQUIREMENTS

Combat missions in the past ten years have proved beyond any doubt that the

dependence on the Unmanned Aerial Vehicle (UAV) has increased significantly. The

demand for UAVs has evolved from conventional hot-war engagements to low-intensity

conflicts such as peacekeeping and humanitarian missions. The demand is so great that it

has outpaced the ability for research and development of UAVs to meet the wide

spectrum of operational requirements.

The UAVs in the current battlefield play a critical role in areas such as

information superiority, collateral damage, urban area fighting and precision strikes

against high payoff targets. The three main areas that have been identified for future

UAV developments include the growth in size of strategic UAVs for greater payload and

endurance, reduction in size of tactical UAVs and the addition of weapons to UAVs to

offer lethal capabilities in combat missions [1].

The desire of shorter “kill-chain”—the time from when the target is spotted by a

sensor until a shooter locks on—is the main force driving the evolution of the Unmanned

Combat Aerial Vehicle (UCAV). The current trend towards asymmetrical warfare, where

90% of the targets are mobile or of the “emerging” variety, emphasizes this even more.

The effects of a short “kill-chain” are less apparent at the operational and strategic levels,

while the values of rapid targeting are obvious at the tactical level of war. If the process is

slow, time critical opportunities may be lost or mobile targets may be long gone by the

time an air strike arrives. The faster the attack, the less time the enemy has to react, adjust,

adapt, mount a counter-offensive or escape [2].

This thesis aims to provide a low-cost solution through integrating commercial

off-the-shelf (COTS) technologies to produce a tactical hunter-killer system that satisfies

the needs for both the reduction in size of tactical UAVs and the weaponization of UAVs

to offer lethal capability.

2

B. PROPOSED CONCEPT OF OPERATIONS

The proposed tactical hunter-killer system is better known as Tactical Unmanned

Combat Aerial Vehicle or Tactical UCAV. Currently, due to high cost and limited

quantities, UCAVs are only made available to military planners at the operational or

strategic level. Low-cost production through integrating commercial off-the-shelf

technologies will be the driver for making such superior war-fighting capabilities

accessible to ground units in the near future.

The Tactical UCAV is designed to enhance the information collection and

autonomous precision strike capability of the ground units. The Tactical UCAV can also

be deployed as sensor nodes as part of a larger global information grid in a network-

centric warfare operation. The proposed concept of operations for the Tactical UCAV is

shown in Figure 1 and can be summarized as follows:

A Hunter Unmanned Aerial Vehicle (HUAV) carries high-resolution sensors and

communication devices. As the name implies, the HUAV will predominantly play the

hunter roll in the proposed Tactical UCAV system. She is configurable to carry one or

many of Killer Unmanned Aerial Vehicles (KUAV), and in this aspect she can be thought

of as the “Mother Ship” or the air command center for the KUAVs.

The KUAV can be configured to carry a payload of a sensor, a warhead or both,

and it will predominantly play the roll of the killer in the proposed system.

After the target is located by the hunter, the information about the target will be

“transferred” to the killer. The killer will then be autonomously launched from the mother

ship. The killer vehicle will autonomously maneuver to the target with the help of

constant updates from the mother ship as to the position of the target.

3

Figure 1. Proposed concept of operations

The Tactical UCAV system is designed to be configurable and low cost, so it can

be deployed as a single system or in numbers (swarm) to fulfill many operation

requirements. The possibilities include:

Single System

Hunt and Kill Mission. As the name implies, upon acquiring a high-payoff target

by the hunter, a killer will be launched to destroy the target.

Continuous Surveillance Mission. Instead of destroying the target, the “killer”

carrying surveillance equipment can be launched to close in or follow the target for

continuous surveillance.

Multiple Systems (Swarm)

Hunt and Kill Mission. This is similar to the first single system mission.

However, it would be used when multiple targets, such as a column of tanks, are to be

engaged simultaneously. Multiple Tactical UCAVs can be assembled in the area of

operation. If each launches two killer vehicles, a swarm of killers is now available to

simultaneously engage the targets.

PVNT Workstation

Piccolo Ground Station

ProcerusGround Station

Killer

Target

Hunter

ATT Computer and NPS Ground Station

Existing NPS System Thesis Research Area

Future Integration

PVNT Workstation



Killer

Target

HunterHunter


Existing NPS System Thesis Research Area

Future Integration

4

Continuous Surveillance Mission. A single system has a limited area of coverage

in any surveillance mission. A swarm of sensors forming a sensor grid can be deployed to

cover a large area or track a moving target with greater tracking accuracy. This operation

can also be part of the global information or sensor grids in a network-centric warfare

operation.

C. SCOPE AND OBJECTIVES

It can be seen from Figure 1 that this thesis will focus on the development of a

prototype KUAV. Due to the limited duration available for this thesis, the integration of

the prototype KUAV with the existing HUAV “Rascal” [3] developed and operated by

the Naval Postgraduate School (NPS) will be done in the near future. The combination of

the prototype KUAV with the existing HUAV will form the Tactical UCAV system.

The scope of this thesis includes the following:

Development of Prototype

Design and build the test prototype

Aerodynamic performance analysis with linear panel software

Tradeoff studies for various performance parameters

Guidance and control analysis with numerical methods

Definition of engagement envelope

Hardware-in-the-Loop Testing

Architecture and setup

Procedures for working with Procerus Virtual Cockpit 2.4.2

Refining engagement envelope

D. THESIS ORGANIZATION

Chapter II provides background on the UCAV and Tactical UAV systems

currently in service or development. Chapter III presents the critical considerations and

5

decisions made for the design and building of the prototype. In Chapter IV, the

methodology and tools used for generating aerodynamics performance data from linear

panel codes will be discussed. Subsequently, the results of the tradeoff studies for key

performance parameters of the prototype will be presented. In Chapter V, the

methodology and simulations used for modeling the practical operating environment with

numerical techniques will be discussed. The results of the simulations will also be

presented in this chapter. In Chapter VI, the setup, procedures, results and findings

obtained from Hardware-in-the-Loop (HIL) experiments will be discussed. Finally in

Chapter VII, the conclusion and recommendations for future works will be discussed.

6


7

II. LITERATURE REVIEW

A. EXISTING UCAV SYSTEMS

Major UAV systems have been well developed over the years. In recent years,

much work and effort have been invested in the development of hybrid systems to reduce

the time required in the sensor-to-shooter targeting cycle. These hybrid systems primarily

integrate the sensor and shooter in a single platform more commonly known as

Unmanned Combat Air Vehicles (UCAV). As part of the literature review, two in-service

UCAVs are presented.

1. Predator B – General Atomics Aeronautical Systems, USA

The Predator B aircraft was developed by General Atomics Aeronautical Systems,

USA, in 2000 and its flight commencing in February 2001. Powered by a turboprop

engine, the Predator B series was designed as a long endurance, high-altitude unmanned

aircraft for use as a multi-mission system. From reconnaissance, surveillance, targeting

and weapons delivery to scientific research and other civilian applications, Predator B has

the capacity to conduct multiple missions simultaneously due to its large internal and

external payload capacity. The detailed technical specifications are presented in

Appendix A. [4]

Figure 2. Predator B armed with Hellfire missiles

8

2. Sperwer B – Sagem, France

Sperwer B was designed and developed by Sagem, France, to support Intelligence,

Surveillance, Target Acquisition and Reconnaissance (ISTAR) at the battlegroup level.

Sperwer is capable of carrying multiple payloads ranging from Forward Looking Infra-

Red (FLIR), Electro Optic / Infra-Red (EO/IR) and Synthetic Aperture Radar (SAR). It is

equipped with two underwing hardpoints that can carry external loads of up to 30kg each.

Sagem has already demonstrated the integration of Sperwer B with the Spike LR missile.

They are cooperating with GIAT, to test a new smart munition delivery system, based on

the Bonus submunition. However, the armed configuration requires that up to 20 kg of

fuel must be removed. This will limit the endurance of the armed Sperwer. The detailed

technical specifications are presented in Appendix B. [5]

Figure 3. Sperwer B armed with Spike LR missiles

B. EXISTING TACTICAL UAV SYSTEMS

The development of the Tactical UAV has also progressed significantly in recent

years. They are evolving into multi-role, multi-mission platforms. As UAV technology

matures, UAV will become increasingly cost effective, they will grow smaller and be

able to accomplish a greater number of missions and fill a greater number of roles.

Besides their current applications in Reconnaissance, Surveillance and Target Acquisition

(RSTA), the Tactical UAV mission set could be expanded to include target designation,

9

strike, mine countermeasures, electronic warfare and information warfare. Also as part of

the literature review, two in-service tactical UCAVs will be presented.

1. Shadow 200 – AAI Corporation, USA

The Shadow 200 short-range Tactical UAV was developed by AAI Corp. in the

1990s. It has a non-retractable tricycle landing gear for conventional wheeled take-off

and landing. The RQ-7A can also be launched from a catapult and has a tailhook to catch

arresting cables for a shorter landing strip.

Shadow 200 is capable of spotting targets up to 125 kilometers away from the

brigade tactical operations center. It has both day and night capability of identifying

tactical vehicles up to 8,000 feet above ground at a slant range of more than 3.5

kilometers. The Shadow ground control station provides near real-time targeting data for

precision weapons using “leap ahead” technology and transmits near real time imagery

and telemetry data to the end users. The detailed technical specifications are presented in

Appendix C. [6]

Figure 4. Shadow 200 being launched from a catapult

10

2. Watchkeeper Tactical UAV – Thales, UK

The Watchkeeper Tactical UAV system will provide the UK armed forces with

Intelligence, Surveillance, Target Acquisition and Reconnaissance (ISTAR) capability.

Watchkeeper is a tactical system that will be operated on the battlefield by the British

Army Royal Artillery. The air vehicle will be capable of carrying a range of sensors

including day and night cameras and surveillance radars. Two WK450 air vehicles will

be able to operate in tandem, with the second acting as a communications relay. The

ground control station will be network enabled to ensure comprehensive communications

links to, for example, airborne stand-off radar, attack aircraft and battlegroup

headquarters. The Watchkeeper system will enter service in the British Armed Forces

Royal Artillery in 2010. The detailed technical specifications are presented in Appendix

D. [7]

Figure 5. Watchkeeper Tactical UAV

11

III. BUILDING THE PROTOTYPE

A. PROTOTYPE DESIGN

After much research and many reviews have been done on open-source data, it is

apparent that there is no commercially available system that can meet the requirements of

the proposed KUAV. Therefore, part of this thesis involved the designing and building of

the prototype that is capable of integrating with “Rascal” for concept demonstration.

The proposed concept of operations placed many constraints on the design of the

KUAV. The three principal considerations for the KUAV are as follows:

Size and weight—It must be small and light so that it can easily integrate with

“Rascal”

Lift and drag—It must have high lift and low drag in order to increase the payload

and reduce the thrust requirements

Noise signature—The noise signature must be kept to a minimum so that it can

have a stealthy approach

With the principal considerations in mind, a non-powered glider with swept

linearly tapered wing has been selected as the design for research and experimentation in

this thesis.

Due to the tight time schedule, a quick study was conducted on in-service glider

systems currently available. The decision was made to design the prototype by scaling

down (maintaining an aspect ratio of AR = 2.78) the “Swift 2 Wing” hobby flying wing

manufactured by Northeast Sailplane Products [8]. Apart from being able to rapidly

produce the prototype, the selection of the autopilot was the other important factor.

KESTRELTM Autopilot 2.2, manufactured by Procerus Technologies, [9] will be used to

control the autonomous flight of the KUAV. This autopilot is being used on a similar

12

class of test platforms that have tapered wings that are swept linearly. Therefore, this

could potentially save much effort and time on tuning the Proportional, Integral and

Derivative (PID) gains of the autopilot.

1. Dimensions

Apart from the reasons cited above, the size of the prototype also ensures that it

can be easily carried and released without hindering the normal operations of “Rascal.”

EH 3.0/12 (3% camber and 12% thick) airfoils, designed by [10], were selected. Table 1

shows the detailed dimensions. Figure 6 shows the plan view of the prototype and the

cross section of the airfoil.

Span Root Chord Taper Tip Chord Sweep Area(mm) (mm) (%) (mm) (deg) (mm2)501.8 203.2 77.6 157.6 34.3 90516

Table 1. Dimensions of prototype

Figure 6. Plan view of prototype and cross section of EH 3.0/12

13

2. Material

The wing was made from commercial off-the-shelf hobby grade Spyder foam.

The foam is durable, lightweight and more importantly, readily available at a low cost. It

is can be easily cut and shaped using a heated wire. The two main disadvantages of

cutting this material (by amateurs) are the surface roughness and the difficulty of shaping

a consistent and accurate airfoil shape. They both will decrease lift and increase parasite

drag on the airfoil, thus reducing the performance of the airfoil. With these disadvantages

in mind, this is still the best solution for rapid prototyping. The elevons of the prototype

were made of lightweight balsa wood and the winglets were made of 1/32” aircraft ply.

3. Components and Loading

The prototype was first tested for its aerodynamic performance using a Remote

Controlled (RC) flight mode. Only after successful RC flights was the prototype tested

for autonomous flight using the Auto Pilot (AP). Different commercial off-the-shelf,

hobby-grade components were required for each of these individual flight modes. A

detailed breakdown of the components and the respective weight for both modes are

presented in Appendix E.

4. Trimming and Balancing

The theoretical aerodynamic center of the prototype was estimated to be at 25%

of the Mean Aerodynamic Chord (MAC) computed using equation (1).

( )( )

2123 1

t tMAC rc

t

+ += × ×

+ (1)

where rc Root Chord and t Taper Ratio= =

MAC was computed to be 181 mm and the theoretical aerodynamic center is

located at 128 mm on the root chord from the nose. After balancing all the components

and trimming from hand-launched RC test flights, the optimum center of gravity CG was

14

found to be located at approximately 118 mm from the nose on the root chord resulting in

a static margin of 10 mm. The details are depicted in Figure 7.

Figure 7. Location of aerodynamic center and CG

Since this static margin is already sufficient for trim flight during hand-launched

test flights (low altitude and short duration). The author is confident that the prototype

KUAV will exhibit more stable performances during “Rascal” launched test flights (high

altitude and long duration).

5. Completed Prototype

The main components (from left to right: autopilot, battery-pack, GPS and

modem) used for the autonomous flight are shown in Figure 8.

Figure 8. Components for autonomous flight

MAC

25% of MAC

Aerodynamic Center

Center of Gravity

15

The prototype KUAV is shown Figure 9. The components described above except

the GPS, will be packed in an opening cut on the airfoil under the plastic cover shown in

Figure 9. A separate slot will be cut on the airfoil to house the GPS unit.

Figure 9. Side and top views of the prototype KUAV

16


17

IV. AERODYNAMICS STUDIES

A. BASIC AERODYNAMICS FOR GLIDERS

The aerodynamic forces acting on a non-powered glider closely resemble those on

any aircraft. The key distinction is that the non-powered glider does not produce thrust

force to counteract the drag force. As a result, a glider cannot maintain a level flight at

constant speed and altitude. A glider in steady gliding flight is always descending relative

to the air around it. The glider is essentially trading altitude to maintain its velocity.

Figure 10 shows the balance of forces acting on a glider descending in a steady glide at

constant velocity.

Figure 10. Forces acting on a glider in flight

The glider follows a path that slopes at an angle below horizontal called the Glide

Angle, denoted by the Greek letter (γ). The direction of the glider’s motion along the

glide path is shown by the velocity arrow on the diagram. (Note that the glider’s

longitudinal axis is not pointing along the direction of motion, but slightly above the

glide path at angle α, called the Angle of Attack, AOA).

Drag (D)

Velocity (V) – Flight Path

Horizontal Lift (L)

Glide Angle (γ)

Angle of Attack (α) γ

WD

WL

18

The lift force (L), which according to linear theory acts at a right angle to the

velocity through the air, is tilted forward from vertical at an angle equal to γ. Drag (D)

acts directly opposite the direction of motion through the air.

At trim condition (constant glide angle and velocity) the following relationships

can be derived.

cosLL W L mg γ= → = (2)

sinDD W D mg γ= → = (3)

It can be observed that when all forces on the glider are balanced there is no net

force to cause acceleration, so the glider will move along the glide slope at a constant

velocity. The velocity (V) and glide angle (γ) can be derived from equations (2) and (3)

respectively.

2 21 12 2L DL AV C and D AV Cρ ρ= =

21 cos2 LAV C mgρ γ∴ = (4)

21 sin2 DAV C mgρ γ∴ = (5)

1 1tan tan tanD D

L L

C C DC C L

γ γ γ− −= → = → = (6)

( )2

cos sinL D

mgVA C Cρ γ γ

=+

(7)

B. SIMULATION USING ANALYTICAL TECHNIQUES

Commercial off-the-shelf software, LinAir (Version 4.3) [11] was used in this

thesis to analyze the aerodynamic characteristics of the glider. LinAir is a linear panel

code program capable of computing aerodynamic characteristics of multi-element,

nonplanar lifting surfaces for a given wing geometry and angle of incidence. The results

19

generated are based on solving the Prandtl Glauert linear partial differential

equations for inviscid, irrotational and subsonic flow.

1. Verification of Panel Codes

There is a need to first “calibrate” several input parameters to LinAir, so that the

results generated can be benchmarked with existing wind tunnel results for a similar class

of wing profile. This is especially true for cambered airfoil, as the procedures for

modeling airfoil with camber stated in [11] are strictly adhered to. It was found that

EH3.0/12 has zero lift at AOA of approximately -1°. Therefore, all results generated will

be adjusted by adding -1° to the actual incidence angle.

In the proposed operating envelope of the glider (low altitude and low speed) the

Reynolds Number (Re) is expected to fall in the region of 50,000 to 100,000. Wind tunnel

experimental results obtained by [12] provide a good insight into the aerodynamic

performance of low aspect ratio wings at low Re. This served as an excellent independent

source to cross-check the data generated by LinAir.

Many aerodynamic parameters, in particular those related to drag, cannot be

computed analytically or by linear methods and are even difficult to get right using

Navier-Stokes solvers, and are usually obtained from wind tunnel experiments. Extensive

tests for wings of AR = 1 and 2 at Re = 70,000 and 100,000 have been conducted and the

results are presented in [12]. In the absence of wind tunnel facilities and time, the

following parameters, Drag Coefficient at Zero Lift (0DC ) and Induced Drag Coefficient

( K ) for the prototype glider (AR = 2.78) will be inferred based on the results published

by [12].

It is assumed that the total drag produced by a wing is give by equation (8). The

values of 0DC and K can be obtained by plotting DC vs. 2

LC for each model and applying

a least-square linear regression to the data. The wind tunnel data obtained by [12] are

summarized in Table 2.

0

2D D LC C KC= + (8)

20

AR CD0 K1 0.015 0.52 0.020 0.4

Table 2. Wind tunnel results of 0DC and K for wings of AR 1 and 2

These data were subsequently run on LinAir using wing profiles matching the

respective AR. The results generated for AR = 2 (AOA from 0° to 10°) closely matched

those obtained in the wind tunnel. However, for AR = 1, LinAir produced a better match

using 0.4K = . The detailed results are presented in Appendix F.

Since the prototype has an AR of 2.78, by extrapolating from the above simple

verification, 0

0.022DC = and 0.4K = will be used to model the prototype glider. LinAir

has proven to be suitable and will be used as the analytical tool for estimating the

aerodynamic performance of the prototype.

2. Geometry and Input File

The geometry of the glider described in Chapter III was modeled using a total of

40 panels covering the entire wing surface and the two elevons as shown in Figure 11.

The input file is shown in Appendix G.

Figure 11. Geometry of glider generated by LinAir

21

3. Aerodynamic Data

The aerodynamic performance of the glider was evaluated for AOA from 0° to 20°

at an interval of 0.5°. (This range of AOA is only feasible for ideal linear theory

simulations using LinAir. The operating AOA range is expected to be lower in practical

model and physical implementation.) Negative AOA were not examined as it is not

possible to maintain steady flight for negative AOA for a non-powered glider. More

importantly, it does not contribute to any performance improvement. Numerical results

generated by LinAir for both the non-cambered airfoil and with the adjustments made for

the cambered airfoil are shown in Appendix H.

C. ANALYTICAL TRADEOFF STUDIES

Three main performance parameters of the glider were examined.

Glide Velocity (V)

Horizontal Range (R)

Airborne Endurance (E)

These performance parameters are also functions of AOA and γ. Using the data

generated, the following five tradeoff studies were analyzed at a release altitude of 300 m

which is a typical operating ceiling for tactical systems.

Tradeoff Study 1 - L/D vs. AOA

Tradeoff Study 2 – γ vs. L/D

Tradeoff Study 3 – R vs. L/D

Tradeoff Study 4 – V vs. AOA

Tradeoff Study 5 – E vs. V

1. Tradeoff Study 1 – L/D vs. AOA

It is important to recall at this juncture that the values of L used in this analysis

are estimated by LinAir. LinAir is a linear panel code program, it assumes that L always

22

varies linearly for all AOA and does not include the effects of flow separation and stall

which are dominant at higher AOA. Conversely, the effects of flow separation and stall

are approximated in D, as D is a quadratic curve fit to empirical data. (The plot showed in

Figure 12 will change if L is computed accurately with considerations to the effects of

flow separation and stall.)

This study is confined to the region of AOA around 3° to 6° where the effects of

flow separation and stall are negligible. Therefore, the results presented in Figure 12 in

the region of AOA around 3° to 6° are valid to this study.

0 2 4 6 8 10 12 14 16 18 201

1.5

2

2.5

3

3.5

4

4.5

AOA − Angle of Attack (deg)

L/D

− L

ift o

ver

Dra

g R

atio

Tradeoff Between L/D and AOA

Figure 12. Tradeoff between L/D and AOA

Analysis of Figure 12 shows that the maximum L/D of approximately 4.4

occurred at an AOA of 3°. The data presented in Table 3 shows the comparison between

the analytical results to those obtained by [12] through wind tunnel experiments.

23

S/No AR L/D AOA Re1 1.00 4.8 6.0° 70,0002 2.00 5.0 4.0° 70,000

3 2.78 4.4 3.0° 70,000

Experimental Wind Tunel Data

Analytical LinAir Result

Table 3. Comparison between experimental and analytical results

The above comparison shows that the results obtained from LinAir are

comparable to the wind tunnel experiment and the selected values of 0

0.022DC = and

0.4K = are within acceptable engineering estimates.

2. Tradeoff Study 2 – γ vs. L/D

It was shown in Equation (6) that γ is a function of L/D. The greater the L/D, the

gentler the glide slope will be and vice versa. This result is shown in Figure 13.

02

46

810

1214

1618

20

1

1.5

2

2.5

3

3.5

4

4.5

0

5

10

15

20

25

30

35


Tradeoff Between Gramma, L/D and AOA

L/D − Lift over Drag Ratio

Gra

mm

a −

Glid

e A

ngle

(de

g)

Figure 13. Tradeoff between γ, L/D and AOA

24

The graph of Tradeoff Study 1 was projected on the x-y plane and the z-axis

shows the values of γ with varying AOA and L/D.

Therefore, following the results of Tradeoff Study 1 and the relationship

presented in equation (6), γ would first decrease with increasing AOA. After stall

condition, γ would increase with increasing AOA. A minimum glide angle of 12.7° would

occur at the AOA of 3° where L/D was at the maximum.

3. Tradeoff Study 3 – R vs. L/D

Figure 14. Glide path geometry

The glide path shown in Figure 14 can be related through equation (9). The

gentler the glide slope (smaller γ), the farther the glider will travel horizontally before

reaching the ground. Therefore, maximum R will occur at minimum γ.

tantan

H HRR

γγ

= → = (9)

Velocity (V)

Horizontal

α

R

H

γ

γ

25

02

46

810

1214

1618

20

1

1.5

2

2.5

3

3.5

4

4.50

5

10

15

20

25

30

35


Tradeoff Between Gramma, L/D and AOA

L/D − Lift over Drag Ratio

Gra

mm

a −

Glid

e A

ngle

(de

g)

Figure 15. Tradeoff between R, L/D and AOA

The result of maximum R occurring at minimum γ is shown in Figure 15. The

graph of Tradeoff Study 1 was again projected on the x-y plane and the z-axis shows the

values of R with varying AOA and L/D.

Therefore, following the results of Tradeoff Study 1 and 2 and the relationship

presented in equation (9), R would first increase with increasing AOA. After stall

condition, γ would decrease with increasing AOA. A maximum R of approximately 1.335

km would occur at the AOA of 3° where γ is at the minimum and L/D is at the maximum.

4. Tradeoff Study 4 – V vs. AOA

It has been explained that a non-powered glider does not produce thrust to counter

drag. In order to maintain a trimmed glide path, it is always trading off altitude to

maintain its velocity. Equation (7) shows the analytical expression of the glide velocity at

trimmed flight with a fixed AOA.

26

0 2 4 6 8 10 12 14 16 18 204

6

8

10

12

14

16

18

20

22

24


V −

Glid

er V

eloc

ity (

m/s

)

Tradeoff Between V and AOA

Figure 16. Tradeoff between V and AOA

Analysis of Figure 16 shows that V increases exponentially at low AOA and

decreases gradually at high AOA. This behavior of V is critical to the next tradeoff study

where speed is varied to achieve the desired airborne endurance.

5. Tradeoff Study 5 – E vs. V

Trimming the glider to fly at its maximum L/D will allow it to fly farthest

horizontally, but it will not result in the longest airborne endurance (E) (duration of time

in the air).

E depends on the vertical sink rate, or rate of decrease in altitude (H), as shown in

equation (10). This depends on both the glide angle and the glide speed. Maximum E is

achieved by trimming the glider to fly at an AOA higher than that where maximum L/D

occurs. (L/D will decrease slightly from the maximum value.) This will result in a steeper

27

glide slope, but it will cause a significant reduction of glide velocity. The reduction in

speed is sufficient to compensate for the effects of the steeper glide slope that are needed

to achieve maximum E.

sinyy

HE where V VV

γ= = (10)

02

46

810

1214

1618

20

46

810

1214

1618

2022

24

0

20

40

60

80

100

120

140


Tradeoff Between E, V and AOA

V − Glider Velocity (m/s)

E −

Airb

orne

Enu

dran

ce (

s)

Figure 17. Tradeoff between E, V and AOA

Maximum E is achieved by trimming the glider to fly at a higher AOA, and this

result is shown in Figure 17. The graph of Tradeoff Study 4 was projected on the x-y

plane, and the z-axis shows the values of E with varying AOA and V.

Therefore, following the results of Tradeoff Study 4 and the relationship

presented in equation (10), E would first increase with increasing AOA until slightly after

maximum L/D condition. Following maximum L/D condition, E would decrease with

28

increasing AOA because a decrease in speed is insufficient to compensate for the effects

of the increasing glide angle. A maximum E of approximately 124 s would occur at the

AOA of 6°.

0 2 4 6 8 10 12 14 16 18 20400

500

600

700

800

900

1000

1100

1200

1300

1400


R −

Hor

izon

tal D

ista

nce

(m)

Comparison Between R and E vs AOA

0 2 4 6 8 10 12 14 16 18 2030

40

50

60

70

80

90

100

110

120

130

E −

Airb

orne

End

uran

ce (

s)

Figure 18. Comparison between R and E vs. AOA

Finally, it can be concluded from the analysis of Figure 18 that maximum R

occurs at an AOA of 3° where L/D is the maximum. Maximum E occurs at a slightly

higher AOA of 6° where further reduction in speed is sufficient to compensate for the

effects of the steeper glide slope. Maximum E can only be achieved at the expense of

horizontal distance (approximately 16.5% reduction in R).

The results obtained from the above theoretical tradeoff studies under ideal

operating conditions will define the most optimistic operational envelope of the prototype

KUAV. The tradeoff between critical performance parameters provides a basis for the

end-user to plan for different engagement scenarios like long range (where maximum

range is desired), loiter (where high endurance is desired) and dash (where high speed is

29

desired). In Chapter V, the operational envelope of the prototype KUAV will be

examined under practical operating conditions. These conditions will be modeled as

closely as possible using numerical methods (MATLAB Simulink).

30


31

V. GUIDANCE AND CONTROL

A. SIMULATION USING NUMERICAL TECHNIQUES

The analysis of the aerodynamic performance data generated by LinAir was based

purely on analytical closed-form solutions derived from equations of motions. These

ideal conditions can never be replicated in practical implementations. The actual

performance of the prototype will be dependent on many other factors such as sensor

measurement errors, variations in atmospheric conditions, the performance of the

autopilot and control algorithm and the efficiency of the mechanical devices onboard.

It is almost impossible to simulate the exact operating environment with current

technology, in particular for conditions involving low Re where variations in

aerodynamic behaviors and performances are highly non-linear. However, there is still an

absolute need to account for as many imperfections as possible in order to better estimate

and predict the operational envelope of the prototype KUAV. MATLAB Simulink and

the aerodynamic performance data generated by LinAir will be utilized to model a more

realistic operating environment for the prototype.

This process is summarized in Figure 19.

Figure 19. Flow chart depicting simulation process

Analytical Performance in LinAir

6DOF + AP Model in Simulink

Modeling Prototype in Simulink

Results and Comparisons

Operational Envelope

Simulations to Investigate Operating Conditions for Optimum R, E and V Performance

Analytical Performance in LinAir

6DOF + AP Model in Simulink

Modeling Prototype in Simulink

Results and Comparisons

Operational Envelope

Simulations to Investigate Operating Conditions for Optimum R, E and V Performance

32

1. Analytical Results from LinAir

Chapter III illustrated how the results generated by LinAir were used to

investigate the aerodynamic performance of the prototype based on analytical equations

of motions in ideal operating conditions. These dimensionless coefficients from LinAir

will be used to define the aerodynamic performance of the prototype in the model

described below.

2. 6 Degree of Freedom (DOF) + Auto Pilot (AP) Simulink Model

An existing Simulink Model for Silver Fox UAU previously developed by [13]

has been used to model the performance of the prototype. A straightforward modification

was made to the model to more closely approximate the configuration of the prototype

KUAV. The engine was removed from the model because the non-powered glider does

not produce thrust. The modified model is shown in Figure 20. It is comprised of the

following three main blocks: the AP Inner loop, Forces and Moments and the 6DOF

equations of motion.

Figure 20. Simulink Block Diagram for 6DOF + AP Model

The 6DOF EOM block implements the equations for a constant-mass rigid-body

6DOF aircraft with respect to the body frame {B}. The detailed derivations will not be

presented in this thesis as they can be found in [14]. In summary, the motions of a 6DOF

aircraft can be completely described by the following set of equations:

AP Inner Loop

Forces and Moments 6DOF EOM

33

Linear Momentum Equation

( )B B B BF m v vω= + ×

Angular Momentum Equation

( )B B B BG J Jω ω ω= + ×

Euler Equations relating body rates and attitudes angles

( )sin cos tanp q rϕ ϕ φ θ= + +

cos sinq rθ φ φ= −

( )sec sin cosq rψ θ φ θ= +

The Forces and Moments block implements the forces and moments acting on the

prototype due to aerodynamics and gravitational effects. They can be expressed as:

B BBaero grav

BBaero

F FFGG

⎡ ⎤⎡ ⎤ += ⎢ ⎥⎢ ⎥

⎣ ⎦ ⎣ ⎦

For simplicity, the aerodynamic forces and moments acting on the aircraft are

often defined in terms of dimensionless aerodynamic coefficients that are functions of the

deflections of the control surfaces.

DB

W Y

BLaero

Baero l

BW m

n

CR C

CFqS

G C bR C c

C b

⎡ ⎤⎡ ⎤⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎡ ⎤ ⎣ ⎦⎢ ⎥=⎢ ⎥

⎡ ⎤⎢ ⎥⎣ ⎦⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎢ ⎥⎣ ⎦⎣ ⎦

Where q is the dynamic pressure and S is the wing reference area.

BW R is the transformation matrix that rotates a vector from the wind coordinate

frame to the aircraft body frame.

34

The numerical values for the dimensionless aerodynamics coefficients were

determined from the data generated by LinAir. The detailed explanation will be presented

in the next section.

The gravitational forces acting on the rigid body will not generate moments as

they are assumed to be acting on the center of gravity.

00B B

grav LF Rmg

⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦

BW R is the rotation matrix that rotates a vector from the earth frame to the aircraft

body frame.

In summary, the forces and moments acting on a 6DOF aircraft can be completely

described by the following equations.

00

DB B B

W Y L

L

CF qS R C R

C mg

⎡ ⎤ ⎡ ⎤⎢ ⎥ ⎢ ⎥= +⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦

lB B

W m

n

C bG qS R C c

C b

⎡ ⎤⎢ ⎥= ⎢ ⎥⎢ ⎥⎣ ⎦

Finally, the AP Inner Loop block implements the inner loop of the autopilot

developed by [13].

3. Modeling the Prototype KUAV in the 6DOF + AP Model

There is a need to initialize the 6DOF + AP model at the operating conditions and

performance of the prototype KUAV. This initialization process is critical so that the

model will generate accurate results based on realistic operating conditions and

aerodynamic performance of the prototype KUAV. As mentioned in the previous section,

the dimensionless coefficients obtained from LinAir will be used to model the

aerodynamic performance of the prototype.

35

The initialization script file of the model is shown in Appendix I. The coefficients

of performance under the “Aerodynamic Derivatives” are initialized using the data

generated by LinAir (units of per radian). The following approximating relationships are

used to compute the coefficients:

20 1 2D D L LC C AC A C= + +

Y Y Y rC C C rβ δβ δ= +

0Lq

L L L L L e B

C qC C C C C e

vα α δα α δ= + + + +

( )2l l l a l r lp lrB

bC C C a C r C p C rvβ δ δβ δ δ= + + + +

0mq

m m m m m e B

C qcC C C C C e

vα α δα α δ= + + + +

( )2n n n e n r np nrB

bC C C e C r C p C rvβ δ δβ δ δ= + + + +

Where , ,e a rδ δ δ represent the deflection of the elevator, aileron and rudder respectively.

(As the prototype KUAV is designed to be a flying wing, there is no rudder and the

elevons function as both elevators and ailerons. 0rδ = and e aδ δ= .)

, ,p q r are the angular rates, b is the wing span, c is the mean aerodynamic chord

and α and β represent the angle of attack and the sideslip angle respectively.

4. Simulation Procedures

In all realistic non-linear models, huge amount of transient response is often

expected which is detrimental to the accuracy of the results generated. It is important to

note at this point, that only the results from the steady-state conditions, better known as

trim conditions, will be used for analysis. Therefore, there is a need to minimize the

duration of this transient response. This is achieved by running the simulation with the

36

initial conditions as close to the trim conditions as possible. This will greatly improve the

accuracy of the results obtained from the simulations.

Specifically for this study, it has already been shown that the performance of the

non-powered glider depends on the trim conditions. This is primarily determined by the

glider’s AOA which in turn is controlled by the deflection of the elevators. Therefore, it is

necessary to ensure that the initial conditions, with respect to the body frame, correspond

to trim conditions. The following procedures were adopted for this study:

The Linearization Task function in Simulink Control Design Blockset was used to

obtain the steady-state velocity for every 1° change in elevator deflection. It was found

in this step that steady-state conditions could only be achieved for AOA between 1° and

8°. This operating range of AOA is significantly smaller than those used in Chapter IV.

This is expected as LinAir is programmed based on linear theory (linearized solution does

not diverge) and it will generate results even for an AOA of 90°.

The body velocity in the initialization script file was set to the corresponding trim

conditions of the respective AOA. (This is to minimize the transient response and improve

the accuracy of the simulation results.)

The model was run for AOA from 1° to 8°.

The output from the workspace for Range (x-position), Velocity and Endurance

(time) was saved when the altitude (z-position) reached zero.

The results of the simulations for a release altitude of 300m are attached in

Appendix J.

5. Results and Comparisons

The results obtained from the 6DOF + AP model were compared to the results

found in the ideal operating environment (presented in Chapter III).

37

0 1 2 3 4 5 6 7 80.5

1

1.5

2

2.5

3

3.5

4

4.5


L/D

− L

ift o

ver

Dra

g R

atio

Comparison Between Analytical and 6DOF + AP Models for L/D vs. AOA

Analytical6DOF + AP

Figure 21. Comparison between analytical and 6DOF + AP models for L/D vs. AOA

Analysis of Figure 21 shows that the L/D of the 6DOF + AP model is about 16%

lower than the L/D obtained in the analytical model. This result was expected, as

imperfections in practical environments will reduce the performance of any system.

Maximum L/D = 4.4 at AOA = 3.0° (Analytical model)

Maximum L/D = 3.7 at AOA = 2.8° (6DOF + AP model)

38

0 1 2 3 4 5 6 7 85

10

15

20

25

30

35


V −

Glid

er V

eloc

ity (

m/s

)

Comparison Between Analytical and 6DOF + AP Models for V vs. AOA

Analytical6DOF + AP

Figure 22. Comparison between analytical and 6DOF + AP models for V vs. AOA

Analysis of Figure 22 shows that the change in V with increasing AOA exhibits

the same behavior in the 6DOF + AP model as the analytical model. In the region

(AOA<2.5°) where V decreases exponentially, system imperfections caused the glider to

travel at a higher speed. In the region (AOA>2.5°) where V decreases gradually, system

imperfections caused the glider to travel at a lower speed.

39

0 1 2 3 4 5 6 7 8200

400

600

800

1000

1200

1400


R −

Hor

izon

tal D

ista

nce

(m)

Comparison Between Analytical and 6DOF + AP Models for R vs. AOA

Analytical6DOF + AP

Figure 23. Comparison between analytical and 6DOF + AP models for R vs. AOA

Analysis of Figure 23 shows that the R generated from the 6DOF + AP model is

about 17% lower than the R obtained from the analytical model.

Maximum R = 1.335 km at AOA = 3.0° (Analytical model)

Maximum R = 1.110 m at AOA = 2.8° (6DOF + AP model)

40

0 1 2 3 4 5 6 7 80

20

40

60

80

100

120

140


E −

Airb

orne

End

uran

ce (

s)

Comparison Between Analytical and 6DOF + AP Models for E vs. AOA

Analytical6DOF + AP

Figure 24. Comparison between analytical and 6DOF + AP models for E vs. AOA

Analysis of Figure 24 shows that the practical E is about 10% lower than the E

obtained in an ideal environment.

Maximum E = 124 s at AOA = 6.0° (Analytical model)

Maximum E = 111.1 s at AOA = 4.2° (6DOF + AP model)

6. Refinement of Operational Envelope

It was mentioned in Chapter IV that the results obtained from LinAir predict the

most optimistic (upper bound) operational envelope of the prototype KUAV. The results

obtained from the 6DOF + AP model in this simulation predict the most conservative

(lower bound) operational envelope.

41

The maximum horizontal range where the prototype can travel (with a release

altitude of 300 m) is 1.11 km at an AOA of 2.8°. Alternatively, it can sacrifice range to

achieve a maximum airborne endurance of 111 s at an AOA of 4.2°.

The actual performance of the prototype KUAV should fall within the upper and

lower bound predicted by the analytical and the 6DOF + AP model respectively. Prior to

the physical test flight, it is of great importance to conduct the hardware-in-the-loop

(HIL) experiments to predict the actual performance of the prototype KUAV. The HIL

procedures and results are presented in Chapter VI.

42


43

VI. HARDWARE-IN-THE-LOOP EXPERIMENTS

A. ARCHITECTURE AND SETUP

Due to the tight timeline of this research, there is no means for the author to

develop a new guidance law to provide reference commands to the autopilot inner loop.

This is similar to the AP Inner Loop block described in Chapter V. Therefore, COTS

Kestrel Autopilot is selected as the hardware for autonomous flight. Its default GPS

waypoint navigation algorithm will be utilized to provide guidance reference commands

to the autopilot inner loop.

The next step to perform is hardware-in-the-loop (HIL) experiments using the

physical autopilot hardware. These will verify the results obtained in analytical and

numerical analysis done prior to actual test flight. Kestrel Autopilot and Virtual Cockpit

have a built-in ability to simulate a 6DOF UAV through the use of a 3rd party, open-

source simulator called Aviones. Aviones displays the simulated flight in 3D space,

which allows the user to quickly and easily visualize the flight plans.

This is the first attempt by the team to set up HIL for Kestrel Autopilot. Therefore

it is important that the system architecture, set up and experimentation procedures are

properly documented for knowledge retention.

1. Architecture

There are two options available for setting up the HIL. A brief summary of each

option is presented. The detailed descriptions are found in the Kestrel Users Guide [15].

Architecture 1, which uses a TCP/IP connection as shown in Figure 25.

44

Figure 25. HIL architecture using a TCP/IP connections

In the TCP/IP architecture, Virtual Cockpit communicates with Aviones over a

TCP/IP connection. Aviones passes data (body rates and body velocities) from Virtual

Cockpit to the autopilot over a hardwired serial connection. The practical advantage of

this method is that only one serial port is needed (which is the configuration for most

laptop in the current market). The disadvantage is that the modem needs to be unplugged

from the autopilot and the results obtained will not account for any latency due to data

transmission by the modem.

Architecture 2 which uses a standard serial modem interface as shown in Figure

26.

Figure 26. HIL architecture using a standard serial modem interface

In the serial modem architecture, the autopilot communicates with Virtual

Cockpit over the standard serial modem interface. Aviones communicates with the

autopilot over a serial connection. The advantage of this method is that the modem will

be left connected to the autopilot as per actual operating configuration. The results

45

obtained will give a more accurate representation of the predicted performance. The main

disadvantage of this set up is that the user will be exposed to unnecessary radiation

emitted from the commbox.

Weighing between the advantages and disadvantages of the two options and

hardware availability, Architecture 1 was selected for this research.

2. Setup

In this thesis, the TCP/IP architecture was setup for the HIL experiments. The

details procedures are presented in Appendix K.

Additional lessons learned by the author are also documented for future reference.

When future studies are done, the most current version of Virtual Cockpit

Aviones must be loaded. (At the time of this experiment, the latest software versions

were Virtual Cockpit, version 2.4.2 and Aviones, version 2007-11-1)

When using Virtual Cockpit, version 2.4.2, an update for Microsoft DirectX must

be installed. The update can be downloaded from

http://www.microsoft.com/windows/directx/default.mspx

The most current version of the firmware for the autopilot must be programmed

into the autopilot hardware. (At the time of this experiment, the firmware version was

MA8_2_3.6)

B. TUNING AUTOPILOT PID GAINS

Chapter II mentioned that the Kestrel autopilot and Virtual Cockpit were

developed by Procerus Technologies for the Zagi UAV test platforms. As the prototype

KUAV is much smaller with different aerodynamic performance than the Zagi, there is a

need to retune the PID gains of the autopilot to match the dynamic response of the

prototype KUAV.

46

1. Aerodynamic Performance Parameters

The dimensionless coefficients from LinAir in Chapter V will again be used to

define the aerodynamic performance of the prototype in the flight simulator, Aviones.

The detailed procedures are presented in Appendix L.

2. Autopilot PID Gains

After setting up the model in Aviones, the PID gains of autopilot were tuned

using the HIL simulation. The Ziegler Nichols rules were applied during tuning, with

addition advice from Chapter II of the Kestrel Installation and Configuration Guide [16].

In addition, the author had also found the guidelines provided by Cloud Cap Technology

[17] for tuning Piccolo autopilot to be useful during the process of tuning the gains on

Kestrel autopilot. The final PID gains set on Kestrel autopilot are presented in Appendix

M.

C. TEST SCENARIOS AND PROCEDURES

Given the short duration of this thesis and this being the first attempt to run HIL

for Kestrel Autopilot the experiments were conducted with the following decisions in

mind.

GPS waypoint navigation was used as the outer loop of the autopilot providing

the guidance reference commands.

The PID gains were obtained using all available knowledge the author has on

Virtual Cockpit and Aviones.

The aerodynamic performance parameters of the prototype were obtained from

LinAir.

Therefore, the results might not be a true representation of the expected

performance of the prototype KUAV but the processes will be beneficial for future

works.

47

1. Test Scenarios and Procedures

A set of four waypoints (numbered 1 to 4) were set as the test route for the

experiment. Waypoints 1 and 2 are set as intermediate positions for the UAV to reach the

desired test altitude and speed. Waypoints 3 and 4 are on the actual test route, and

waypoint 3 simulates the position where the KUAV is released from the HUAV.

Waypoint 4 simulates the target position. Different release altitudes and horizontal ranges

are achieved by adjusting the height and northing difference between waypoints 3 and 4

respectively.

A series of 7 runs, with velocities ranging from 6 m/s to 16 m/s at an interval of 2

m/s, were conducted. (This range of test velocities is selected based on the results

obtained from both the analytical and 6DOF + AP models in the previous chapters.) The

detailed procedures are presented in Appendix N.

D. RESULTS AND COMPARISON

Using the procedures described in Appendix N, six simulations were completed

and the result of a sample case (velocity of 8 m/s) is presented below.

1. Results

Using the data generated from the HIL for a glide velocity of 8 m/s, the following

graphs are plotted and shown in Figure 27.

The graph of H vs. R, provides insights to the L/D, glide angle and maximum

horizontal range traveled by the KUAV during its descend from 300 m. (Typical

operating ceiling for tactical systems.)

The graph of P vs. E, provides insights to the airborne endurance of the KUAV. It

also shows the amount of transient the KUAU experiences after it is released from the

HUAV. The average pitch angle will also be used to determine the AOA of the KUAV

during its descend. The jump in pitch angle at the terminal phase is to be ignored as the

KUAV had already landed on ground.

48

The graph of V vs. E, provides insights to the average velocity of the KUAV

during its descend. It can also be observed from this graph that there is a transient period

before steady-state condition is reached. The jump in velocity at the terminal phase is

also to be ignored as the KUAV had already landed on ground.

200 400 600 800 1000 1200 14000

100

200

300

R − Horizontal Range (m)

H −

Alti

tude

(m

) H vs. R

0 50 100 150 200 250−15

−10

−5

0

E − Endurance (sec) P −

Pitc

h A

ngle

(de

g) P vs. E

0 50 100 150 200 2507

8

9

E − Endurance (sec)

V −

Vel

ocity

(m

/s) V vs. E

Figure 27. HIL results for a gliding velocity of 8m/s

The results for all the six simulations are summarized and attached in Appendix

O.

2. Comparison

The results obtained from the HIL experiments were compared to those obtained

from the analytical and the 6DOF + AP models. This comparison was only made for a

small AOA range (until 4.5°) as this is the hardware’s physical limits during actual

implementation.

49

0 0.5 1 1.5 2 2.5 3 3.5 4 4.50

1

2

3

4

5

6


L/D

− L

ift o

ver

Dra

g R

atio

Comparison Between Analytical, 6DOF + AP and HIL Models for L/D vs. AOA

Analytical6DOF + APHIL

Figure 28. Comparison between analytical, 6DOF + AP and HIL models for L/D vs.

AOA

Analysis of Figure 28 shows that the L/D computed from the HIL experiments

displayed a similar trend but at a larger value as compared to those obtained in the

analytical and 6DOF + AP models. In addition, the L/D value did not display a distinct

point of maximum within this range of AOA. With the results from all 3 models

displaying a similar trend, this further concludes the expected change of L/D with respect

to changing AOA.

50

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5200

400

600

800

1000

1200

1400

1600

1800


R −

Hor

izon

tal D

ista

nce

(m)

Comparison Between Analytical, 6DOF + AP and HIL Models for R vs. AOA


Figure 29. Comparison between analytical, 6DOF + AP and HIL models for R vs.

AOA

Analysis of Figure 29 shows that the R computed from the HIL experiments

displayed a similar trend but at a larger value as compared to those obtained in the

analytical and 6DOF + AP models. This result is expected, as it is shown in Figure 27

that the values of L/D for HIL are larger. Similarly, the R value did not display a distinct

point of maximum within this range of AOA. With the results from all 3 models

displaying a similar trend, this further concludes the expected change of R with respect to

changing AOA.

51

0 0.5 1 1.5 2 2.5 3 3.5 4 4.50

50

100

150

200

250

300


E −

Airb

orne

End

uran

ce (

s)

Comparison Between Analytical, 6DOF + AP and HIL Models for E vs. AOA


Figure 30. Comparison between analytical, 6DOF + AP and HIL models for E vs.

AOA

Analysis of Figure 30 shows that the E computed from the HIL experiments

displayed a similar trend as compared to those obtained in the analytical and 6DOF + AP

models. However, at higher AOA where the velocity of the glider is lower, the E obtained

from HIL is significantly larger. This is clearly different from the expected decrease in E

with increasing AOA as predicted from the analytical and 6DOF + AP models.

The above results showed that the HIL model generates the same trend and

behavior for E and R with changing AOA as compared to the analytical and the 6DOF +

AP models. However, the results obtained from the HIL experiments were better than

expected. Ideally, the results obtained from the HIL experiments should fall in between

the upper and lower bound predicted by the analytical and the 6DOF + AP models

respectively. This shows that the simulation model used by Kestrel is more optimistic

than of the analytical model. There is a need for future studies to better understand the

control and guidance architecture used in Kestrel autopilot. Apart from the uncertainty

52

caused by the autopilot, the aerodynamics parameters derived from LinAir might not be

the actual representation of the KUAV’s performance. The performance data will have to

be refined after physical flight test. A detailed list of recommended future studies is

presented in the next chapter.

53

VII. CONCLUSIONS

A. CONCLUSIONS

This thesis presented the concept of integrating COTS technologies to produce a

prototype of a Tactical UCAV system. There is an existing HUAV developed and

operated by the Naval Postgraduate School (NPS). For this reason, this thesis focused its

effort on the design and evaluation of the performance of a non-powered glider with a

wing that is linearly tapered and swept as the prototype KUAV.

As there was no readily available model in the market that could fulfill the

requirements of the proposed KUAV, a prototype was physically built for

experimentation. The prototype was made purely from COTS hobby-grade material and

components. The prototype was trimmed for low-altitude, hand-launched RC flight.

The aerodynamic performances of the proposed prototype were first obtained

using commercial panel code software. The aerodynamic performances were

subsequently used in a Simulink 6DOF + AP software simulation, and the results

obtained were comparable to those obtained from theoretical analysis.

HIL simulations were conducted using the default setting in Kestrel Autopilot.

This was the first attempt to run HIL testing for the Kestrel Autopilot. One objective of

running the HIL was to validate the performance of the prototype prior to a high-altitude

autonomous AP test flight. The other main objective was to establish the correct system

architecture and procedures for future studies.

Despite the short duration of this thesis, it shows that the proposed “Tactical

UCAV” can be realized using COTS technologies. The following section describes some

of the future studies equired to reach the envisaged end product.

B. RECOMMENDATIONS AND FUTURE STUDIES

The following future studies are recommended for developing the proposed

“Tactical UCAV” system:

54

1. Understanding the Kestrel Autopilot

There is a need to further understand the performance of the Kestrel autopilot.

Given the short duration of this thesis, the current knowledge of the autopilot is

superficial and insufficient to explain certain results or behaviors obtained from the HIL

simulations.

2. Variations to Gliding Profile

This thesis had focused the research on investigating the maximum horizontal

distance and airborne endurance the prototype can achieved given a released altitude.

Future works should investigate the performance of the prototype for different glide

profiles, for example the target is located right below the KUAV and not at a distance

away.

3. Physical Integration with “Rascal”

“Rascal” is an SUAV developed and operated by the Naval Postgraduate School

(NPS), and it will be the HUAV carrying the KUAV. There is a need to develop a

mechanical device to carry and release the prototype from “Rascal”. Only with this

device can the prototype be tested with high-altitude launches.

4. Data Integration with “Rascal”

Apart from developing the mechanical device to integrate the prototype with

“Rascal,” there is also a need to establish an integrated method of communications for

data transfers and updates between the HUAV and the KUAV. Two possible options are

presented below for future works.

55

Option 1: Data transfers and updates are sent from the HUAV to the KUAV via

Piccolo and Procerus ground stations as shown in Figure 31.

Figure 31. Communication protocol via ground control stations

Option 2: Data transfers and updates are sent directly from the HUAV to the

KUAV via XBee (Radio Frequency Module developed by MaxStream, Inc.) as shown in

Figure 32. This option is preferred as the prototype KUAV need not have connectivity

with the Procerus ground station. Effectively, the Procerus ground station can be

eliminated and target data are relayed to the prototype KUAV through the Hunter.

Figure 32. Communication protocol via radio frequency module

PVNT Workstation



Killer

Target

Hunter

Existing CommsStructure

Continuous tracking of the target by the Hunter after the release of the Killer.

Target updates broadcast to PerceptiVU

Communication Protocol to link both ground stations will be developed

Target updates relayed from Procerus Ground Station to the Killer

The Killer glides to the target based on the updated target coordinates


PVNT Workstation



Killer

Target

Hunter










Killer

Target

HunterHunter









Killer

Target

Hunter




Communication Protocol to link the hunter and the killer via XBee will be developed



PVNT Workstation

ProcerusGround Station OptionalPiccolo

Ground Station

Killer

Target

HunterHunter




Communication Protocol to link the hunter and the killer via XBee will be developed



PVNT Workstation

ProcerusGround Station Optional

56

5. Variable Payload Capability for the KUAV

The current aerodynamic performances of the prototype were obtained based on a

fixed payload. The effects of variable payloads should be investigated, as the proposed

KUAV is capable of carry variable payloads.

57

APPENDIX A. TECHNICAL SPECIFICATIONS OF PREDATOR B

58


59

APPENDIX B. TECHNICAL SPECIFICATIONS OF SPERWER B

60


61

APPENDIX C. TECHNICAL SPECIFICATIONS OF SHADOW 200

62


63

APPENDIX D. TECHNICAL SPECIFICATIONS OF WATCHKEEPER TACTICAL UAV

64


65

APPENDIX E. COMPONENT SPECIFICATIONS

A. RC Mode

S/No Items Weight Remarksgrams

1 2 x Servo 102 Radio Receiver 63 Battery Pack 48.84 Power Converter 65 Nose Weight 6.9 Lead

77.7

Structural Weight 67.8 gramsTotal Weight 145.5 grams

Effective Wing Area 71,438 mm2

Average Wing Loading 2.04E-03 grams/mm2

RC Mode

Total

B. AP Mode

S/No Items Weight Remarksgrams

1 2 x Servo 102 Auto Pilot 183 Modem 13.84 GPS Receiver 13.75 Battery Pack 33.26 Power Converter 67 Cables 58 Nose Weight 6.9 Lead

106.6

Structural Weight 67.8 gramsTotal Weight 174.4 grams

Effective Wing Area 71,438 mm2

Average Wing Loading 2.44E-03 grams/mm2

AP Mode

Total

66


67

APPENDIX F. VERIFICATION OF PANEL CODES

A. AR = 1 at Re = 70,000 1. LinAir Input File (CD0 = 0.015 and K = 0.4)

68

2. Results generated by LinAir

0 1 2 3 4 5 6 7 8 9 100.015

0.02

0.025

0.03

0.035

0.04

0.045

0.05

0.055

0.06

0.065

AOA - Angle of Attack (deg)

CD

- C

oeffi

cien

t of D

rag

CD vs AOA for AR = 1 and Re = 70,000

The above results, generated by LinAir for the AOA range of 0° to 10°, closely matched the results obtained in the wind tunnel experiments conducted by [12].

69

B. AR = 2 at Re = 70,000 1. LinAir Input File (CD0 = 0.020 and K = 0.4)

70

2. Results generated by LinAir

0 1 2 3 4 5 6 7 8 9 100.02

0.04

0.06

0.08

0.1

0.12

0.14

AOA - Angle of Attack (deg)

CD

- C

oeffi

cien

t of D

rag

CD vs AOA for AR = 2 and Re = 70,000

The above results, generated by LinAir for the AOA range of 0° to 10°, closely matched the results obtained in the wind tunnel experiments conducted by [12].

71

APPENDIX G. INPUT FILE FOR PANEL CODE

72


73

APPENDIX H. DATA FILE GENERATED BY PANEL CODE

Results for non-cambered airfoil

74

Results for cambered airfoil

Column 1 = Angle of Attack Column 2 = Angle of Sideslip Column 3 = Free Stream Mach Number Column 4 = Lift Coefficient Column 5 = Drag Coefficient Column 6 = Sideforce Coefficient Column 7 = Moment Coefficient Column 8 = Rolling Moment Coefficient Column 9 = Yawing Moment Coefficient Column 10 = Span Efficiency

75

APPENDIX I. INITIALIZATION FILE FOR 6DOF + AP MODEL

% Initialization file for 6DOF + AP simulation model. clc; T=0.1; r_lim=7; % Initial Conditions in ENU (all vector data is represented as a column vectors) Pos_0 = [0; 0; 300]'; % Initial position vector (m) Euler_0 = [0; 0; 0]'*pi/180; % Initial Euler angles (rad) Omega_0 = [0; 0; 0]'; % Initial Omega (rad/s) PQR_0 = [0; 0; 0]'; % Initial Omega (rad/s) Vb_0 = [5; 0; 0]'; % Initial body-velocity vector (m/s) % Mass and Geometric Parameters recomputation S = (0.2032+0.1576)*0.5018/2; % Surface area of wing (m2) span = 0.5018; % wingspan (m) chord = 0.1814; % Mean Aerodynamic Chord (m) mass = 0.175; % gross weight (kg) Ixx = 0.0028959; % main moment of inertia around axis Ox (kg*sq.m) Iyy = 0.0006514; % main moment of inertia around axis Oy (kg*sq.m) Izz = 0.0035430; % main moment of inertia around axis Oz (kg*sq.m) % Aerodynamic Derivatives (all per radian) CL0 = 0.05147; % lift coefficient at a = 0 CLa = 2.9095; % lift curve slope CLa_dot = 0; % lift due to angle of attack rate CLq = 0; % lift due to pitch rate CLDe = -1.3776; % lift due to elevator CD0 = 0.0220; % drag coefficient at a = 0 Apolar = 0.5229; % drag curve slope (A2) CD=CD0+A2*1.25*CL^2 A1 = 0; CYb = 0; % side force due to sideslip CYDr = 0; % sideforce due to rudder Clb = -0.00172; % dihedral effect = -0.00172 Clp = 0; % roll damping Clr = 0; % roll due to yaw rate ClDa = 0; % roll control power ClDr = 0; % roll due to rudder Cm0 = -0.02463; % pitch moment at a = 0 Cma = -1.4192; % pitch moment due to angle of attack Cma_dot = 0; % pitch moment due to angle of attack rate Cmq = 0; % pitch moment due to pitch rate CmDe = -1.1543; % pitch control power Cnb = 0; % weathercock stability Cnp = 0; % adverse yaw Cnr = 0; % yaw damping CnDa = 0; % aileron adverse yaw CnDr = 0; % yaw control power CLDf = -1.3776; % lift due to flap CmDf = -1.1543; % flap control power

76

% Standard Atmosphere ISA_lapse = .0065; % Lapse rate (degC/m) ISA_hmax = 2000; % Altitude limit (m) ISA_R = 287; % Gas Constant (degK*m*m/s/s) ISA_g = 9.815; % Gravity (m/s/s) ISA_rho0 = 1.225; % Density at sea level (kg/m/m/m) ISA_P0 = 101325; % Sea-level Pressure (N/m/m) ISA_T0 = 289; % Sea-level Temperature (degK)

77

APPENDIX J. SIMULATION RESULTS FROM THE 6DOF + AP MODEL

AOA E R V L/D(deg) (s) (m) (m/s)1.0 15.6 252 32.7 0.842.0 58.0 903 16.1 3.012.7 92.1 1106 12.5 3.692.8 96.2 1110 12.0 3.702.9 99.9 1104 11.5 3.683.0 104.6 1084 10.7 3.613.1 106.1 1069 10.5 3.563.2 107.1 1057 10.3 3.523.5 109.0 1020 9.8 3.404.0 110.5 971 9.2 3.244.1 111.0 924 8.8 3.084.2 111.1 879 8.4 2.934.3 110.8 840 8.0 2.805.0 110.2 786 7.6 2.626.0 107.3 624 6.5 2.087.0 105.5 516 5.7 1.728.0 105.0 461 5.3 1.54

78


79

APPENDIX K. PROCEDURES FOR SETTING UP THE TCP/IP CONNECTION

The following set of procedures is used to set up the TCP/IP connection for the

HIL experiments.

Remove the modem from the autopilot.

Disconnect the programming wire on the programming cable as shown in Figure

33.

Figure 33. Disconnected programming wire

Plug the 5 pin header of the programming cable into the autopilot modem port as

shown in Figure 34. The top cable is the programming cable, and the bottom cable is the

power cable.

Figure 34. Connection for the programming and power cables on autopilot

80

Plug the other end of the programming cable into a free serial port on the

computer that will be running Virtual Cockpit.

Power on the autopilot.

Load Aviones.

Load Virtual Cockpit.

Open the “Edit Agents” window in Virtual Cockpit. (This can be found in the

“Agent Menu”) The right-most column is labeled “HIL” for hardware-in-the-loop.

Check the box next to the agent number of the autopilot as shown in Figure 35.

Figure 35. HIL selection in Virtual Cockpit

Click “OK.”

A dialog box will pop up (see Figure 36) in which the IP address of the computer

running Aviones may be specified. If Aviones is running on the same computer that is

running Virtual Cockpit, click the “Loopback” button and click “OK” to connect to

Aviones. Virtual Cockpit will now attempt to connect to Aviones and add all the selected

agents that had the HIL checkboxes checked.

Figure 36. Connection window for Aviones

81

If connected successfully, the Virtual Cockpit will indicate on its status bar and

Aviones will have added the Agent to its list on the bottom of the window, as shown in

Figure 37.

Figure 37. Agent added in Aviones

The next step is to make sure that Aviones is set up to talk to the autopilot through

the correct comm port. Open up the HIL simulation control dialog in Aviones through

“View Menu” followed by “HIL Sim Ctrl”.

Place the cursor in the Serial Port # box.

Delete the value that is currently in the box using the backspace key. (The box

should turn red)

Enter the serial port number connected to the autopilot.

Click “Enter” – The box should turn white and the word “Open” should be

displayed in the box to the right as shown in Figure 38.

82

Figure 38. HIL simulation control window

With the correct serial port open, data should begin to flow between Aviones,

Virtual Cockpit and the autopilot. Virtual Cockpit should show communications with the

autopilot.

To verify communication between Aviones and the autopilot, the right-most LED

on the autopilot should be blinking rapidly. This indicates that the autopilot is receiving

sensor information from Aviones.

The final display on the user interfaces for Aviones and Virtual Cockpit (if there

is a successful connection) is shown in Figure 39.

83

Figure 39. User interfaces for Aviones and Virtual Cockpit

84


85

APPENDIX L. PROCEDURES FOR UPDATING AERODYNAMIC PERFORMANCE PARAMETERS IN AVIONES

The initial performance parameters loaded in Avoines were for Zagi UAV test

platform. The procedures for updating these performance parameters are as follows.

Locate the initialization text file, “physics_params” in the Aviones folder.

Change the necessary parameters to match the performance of the prototype. The

updated parameters for this thesis are shown in Figure 40.

Figure 40. Aerodynamic performance coefficients for prototype

86

Save the changes. (Do not change the file name)

Run Aviones.

Verify that the parameters had been updated by clicking on “View” followed by

“Physics Parameters” as shown in Figure 41. The values of the parameters can also be

changed from this user interface.

Figure 41. Physics parameter editor window

87

APPENDIX M. PID GAINS FOR PROTOTYPE ON KESTREL AUTOPILOT

The PID gains can be changed from the user interface of Virtual Cockpit. Click

on “Settings” followed by “PID Values”.

The PID gains for the prototype on Kestrel autopilot are shown in Figure 42. Only

the parameters shaded in yellow was changed. The remaining parameters were the default

values used for Zagi UAV test platforms.

Figure 42. Values of PID gains

88


89

APPENDIX N. PROCEDURES FOR RUNNING THE HIL EXPERIMENTS

As mentioned in Chapter VI, this was the first attempt to run HIL for Kestrel

Autopilot. Therefore, this set of procedures is described in detail for the benefit of future

works.

Set up the TCP/IP connection as described in Appendix J.

Create a test route for the respective velocity (6 m/s to 16 m/s at 2 m/s intervals),

with waypoints 3 and 4 having a height difference of 300 m. Upload the test route to the

autopilot using the flight plan control window (click on the red “Upld” tab) of Virtual

Cockpit as shown in Figure 43.

Figure 43. Flight control window of Virtual Cockpit

90

Launch the UAV using the “UAV” tab in Aviones by clicking on “Launch if

necessary” or “Relaunch current UAV” as shown in Figure 44.

Figure 44. Launching the UAV from Aviones

Ensure that the UAV is in “Nav” mode and the autopilot is navigating to

waypoint 1 as shown in Figure 45.

91

Figure 45. UAV in “NAV’ mode

The benefit of using Virtual Cockpit’s default settings for the Zagi UAV test

platform is that the Zagi is a powered glider. It is capable of autonomous flight toward

the active waypoint with its onboard propeller motor.

When it has reach waypoint 3, which simulates the release point of the KUAV, it

is necessary to “turn off” the propeller motor as the KUAV is a non-powered glider. This

can be implemented through Aviones. Under the “View” menu in Aviones, open “HIL

Sim Control” and click on the “Engine Failure” tab as shown in Figure 46. This will

“turn off” the motor, but the autopilot is still navigating waypoints.

92

Figure 46. HIL Sim Control Window of Aviones

Virtual Cockpit has a data logging capability and this is extremely useful in

recording the results for subsequent analysis. The data log can be activated by clicking

“Settings” followed by “Data Logs”. The user interface is shown in Figure 47.

Figure 47. Data log window of Virtual Cockpit

93

Data logging commences moments before engine failure is activated to simulation

the releasing of the KUAV from the HUAV.

Four data are recorded for analysis.

Airspeed, the speed at which the KUAV is descending after being released from

the HUAV.

Altitude, the height of the KUAV during its descend after being released from the

HUAV.

GPS North Position, the horizontal range which the KUAV had traveled during its

descend after being released from the HUAV.

Theta, the pitch angle of the KUAV during its descend after being released from

the HUAV.

When then altitude of the UAV reaches zero, this indicate that the UAV have

landed.

The data for each run are saved as M-scripts and analyzed using Matlab. The

following script is written to analyze the results and plot the respective graphs. %Data analysis for HIL

clc;

test8ms;

data=datalog.samples;

E=data(:,1)./5; P=data(:,3).*180/pi; R=data(:,4); V=data(:,5); H=data(:,6);

glide=(data(100,6)-data(250,6))/(data(100,4)-data(250,4)) glide_deg=glide*180/pi

pitch=mean(data(100:250,3)) pitch_deg=pitch*180/pi

aoa_deg=pitch_deg-glide_deg

Velocity=mean(data(100:250,5))

94

Start_Time=data(1,1); End_Time=data(316,1); Endurance=(End_Time-Start_Time)/5

Start_R=data(1,4); End_R=data(316,4); Range=(End_R-Start_R)

figure (1); hold on;

subplot(3,1,1) plot(R,H); axis equal; xlabel('R - Horizontal Range (m)','fontsize',14); ylabel('H - Altitude (m)','fontsize',14); title('H vs. R','fontsize',16); grid on;

subplot(3,1,2) plot(E,P); xlabel('E - Endurance (sec) ','fontsize',14); ylabel('P - Pitch Angle (deg)','fontsize',14); title('P vs. E','fontsize',16); grid on;

subplot(3,1,3) plot(E,V); xlabel('E - Endurance (sec)','fontsize',14); ylabel('V - Velocity (m/s)','fontsize',14); title('V vs. E','fontsize',16); grid on; print -tiff -depsc 8ms.eps;

hold off;

Repeat the above procedures for the other test velocities.

95

APPENDIX O. SIMULATION RESULTS FROM HIL

Pitch Glide Angle AOA R E Velocity H L/D(deg) (deg) (deg) (m) (s) (m/s) (m)-5.06 -9.75 4.69 1752 288 6.0 300 5.84-8.47 -10.99 2.52 1594 200 8.0 300 5.31

-10.79 -12.29 1.5 1449 144 10.0 300 4.83-12.91 -13.53 0.62 1273 106 12.0 300 4.24-15.07 -15.42 0.35 1122 81 14.0 300 3.74-17.39 -17.51 0.12 977 63 16.2 300 3.26

96


97

LIST OF REFERENCES

[1] J. C. Wang, & Y. S. Chua, (2005). Unmanned Aerial Vehicle: Development Trends & Technology Forecast. DSTA Horizons, 1. Retrieved July 10, 2007, from http://www.dsta.gov.sg/index.php/DSTA-2005-Chapter-2/.

[2] J. T. Correll, (2002). From Sensor to Shooter. Airforce Magazine Online,85.

Retrieved July 10, 2007, from http://www.afa.org/magazine/Feb2002/0202edit.asp.

[3] V. Dobrokhodov, I. Kaminer, K. D. Jones, & R. Ghabcheloo, (2006). Vision

Based Tracking and Motion Estimation For Moving Targets Using Small UAVs. American Institute of Aeronautics and Astronautics. Retrieved July 10, 2007, from http://pdf.aiaa.org/preview/CDReadyMGNC06_1305/PV2006_6606.pdf.

[4] General Atomics Aeronautical Systems. Predator B (2007). [Online Brochure]

Retrieved July 10, 2007, from http://www.ga-asi.com/products/pdf/Predator_B.pdf.

[5] Sagem Défense Sécurité, Sperwer (2006). [Online Brochure] Retrieved July 10,

2007, from http://www.sagem-ds.com/pdf/en/D719.pdf. [6] AAI Corporation, Shadow 200 (2006). [Online Brochure] Retrieved July 10, 2007,

from http://www.aaicorp.com/New/UAS/Shadow200_08-23-07a.pdf. [7] Thales UK, Watchkeeper Tactical UAV System (2006). [Online Brochure]

Retrieved July 10, 2007, from http://www.army-technology.com/projects/watchkeeper/.

[8] Northeast Sailplanes Products, Swift 2 Wing (2007). [Online Brochure] Retrieved

Oct 15, 2007, from http://www.nesail.com/detail.php?productID=5482. [9] Procerus Technologies, KESTRELTM Autopilot 2.2 (2007). [Online Brochure]

Retrieved July 10, 2007, from http://www.procerusuav.com/productsKestrelAutopilot.php.

[10] J. Yost, The EH Airfoils. Retrieved July 10, 2007, from

http://www.b2streamlines.com/EH.html. [11] LinAir Version 3.4 User Manual. (2003) Stanford, CA: Desktop Aeronautics, Inc.

98

[12] G. E. Torres, & T. J. Mueller, (2001). Aerodynamic Characteristics of Low Aspect Ratio Wings at Low Reynolds Number. In T. J. Mueller (Ed.) Fixed and Flapping Wing Aerodynamics for Micro Air Vehicle Application (pp. 115-141). Virginia: American Institute of Aeronautics and Astronautics.

[13] M. I. Lizarraga, (2004). Autonomous Landing System for a UAV (Master’s

Thesis, Naval Postgraduate School, Monterey, CA, 2004). [14] B. W. McCormick, (1995). Aerodynamics Aeronautics and Flight Mechanics.

New York: John Wiley & Sons, Inc. [15] Kestrel User Guide Version 1.7. (2007). Vineyard, UT: Procerus Technologies. [16] Kestrel Installation and Configuration Guide Version 1.7. (2007). Vineyard: UT,

Procerus Technologies. [17] Piccolo Steps to Autonomous Flight Version 1.1. (2003). Hood River: OR, Cloud

Cap Technology.

99

INITIAL DISTRIBUTION LIST

1. Defense Technical Information Center Ft. Belvoir, Virginia

2. Dudley Knox Library Naval Postgraduate School Monterey, California

3. Professor Anthony J. Healey Chairman, Department of Mechanical and Astronautical Engineering (MAE) Naval Postgraduate School Monterey, California

4. Professor Vladimir N. Dobrokhodov Naval Postgraduate School Monterey, California

5. Professor Kevin D. Jones Naval Postgraduate School Monterey, California

6. Professor Yeo Tat Soon Director, Temasek Defence System Institute (TDSI) National University of Singapore Singapore

NAVAL POSTGRADUATE SCHOOL · December 2007 Thesis Advisors: ... UU NSN 7540-01-280-5500 Standard...

Documents

Transcript of NAVAL POSTGRADUATE SCHOOL · December 2007 Thesis Advisors: ... UU NSN 7540-01-280-5500 Standard...